1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
|
// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <map>
#include <memory>
#include <optional>
#include <vector>
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "common/swap.h"
#include "core/hle/service/nvdrv/devices/nvdevice.h"
namespace Tegra {
class MemoryManager;
} // namespace Tegra
namespace Service::Nvidia {
class Module;
}
namespace Service::Nvidia::NvCore {
class Container;
class NvMap;
} // namespace Service::Nvidia::NvCore
namespace Service::Nvidia::Devices {
constexpr u32 DEFAULT_BIG_PAGE_SIZE = 1 << 16;
constexpr u32 DEFAULT_SMALL_PAGE_SIZE = 1 << 12;
class nvmap;
enum class AddressSpaceFlags : u32 {
None = 0x0,
FixedOffset = 0x1,
Remap = 0x100,
};
DECLARE_ENUM_FLAG_OPERATORS(AddressSpaceFlags);
class nvhost_as_gpu final : public nvdevice {
public:
explicit nvhost_as_gpu(Core::System& system_, Module& module, NvCore::Container& core);
~nvhost_as_gpu() override;
NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
std::vector<u8>& output) override;
NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
const std::vector<u8>& inline_input, std::vector<u8>& output) override;
NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input,
std::vector<u8>& output, std::vector<u8>& inline_output) override;
void OnOpen(DeviceFD fd) override;
void OnClose(DeviceFD fd) override;
Kernel::KEvent* QueryEvent(u32 event_id) override;
private:
class BufferMap final {
public:
constexpr BufferMap() = default;
constexpr BufferMap(GPUVAddr start_addr_, std::size_t size_)
: start_addr{start_addr_}, end_addr{start_addr_ + size_} {}
constexpr BufferMap(GPUVAddr start_addr_, std::size_t size_, VAddr cpu_addr_,
bool is_allocated_)
: start_addr{start_addr_}, end_addr{start_addr_ + size_}, cpu_addr{cpu_addr_},
is_allocated{is_allocated_} {}
constexpr VAddr StartAddr() const {
return start_addr;
}
constexpr VAddr EndAddr() const {
return end_addr;
}
constexpr std::size_t Size() const {
return end_addr - start_addr;
}
constexpr VAddr CpuAddr() const {
return cpu_addr;
}
constexpr bool IsAllocated() const {
return is_allocated;
}
private:
GPUVAddr start_addr{};
GPUVAddr end_addr{};
VAddr cpu_addr{};
bool is_allocated{};
};
struct IoctlAllocAsEx {
u32_le flags{}; // usually passes 1
s32_le as_fd{}; // ignored; passes 0
u32_le big_page_size{};
u32_le reserved{}; // ignored; passes 0
u64_le va_range_start{};
u64_le va_range_end{};
u64_le va_range_split{};
};
static_assert(sizeof(IoctlAllocAsEx) == 40, "IoctlAllocAsEx is incorrect size");
struct IoctlAllocSpace {
u32_le pages{};
u32_le page_size{};
AddressSpaceFlags flags{};
INSERT_PADDING_WORDS(1);
union {
u64_le offset;
u64_le align;
};
};
static_assert(sizeof(IoctlAllocSpace) == 24, "IoctlInitalizeEx is incorrect size");
struct IoctlFreeSpace {
u64_le offset{};
u32_le pages{};
u32_le page_size{};
};
static_assert(sizeof(IoctlFreeSpace) == 16, "IoctlFreeSpace is incorrect size");
struct IoctlRemapEntry {
u16_le flags{};
u16_le kind{};
u32_le nvmap_handle{};
u32_le map_offset{};
u32_le offset{};
u32_le pages{};
};
static_assert(sizeof(IoctlRemapEntry) == 20, "IoctlRemapEntry is incorrect size");
struct IoctlMapBufferEx {
AddressSpaceFlags flags{}; // bit0: fixed_offset, bit2: cacheable
u32_le kind{}; // -1 is default
u32_le nvmap_handle{};
u32_le page_size{}; // 0 means don't care
s64_le buffer_offset{};
u64_le mapping_size{};
s64_le offset{};
};
static_assert(sizeof(IoctlMapBufferEx) == 40, "IoctlMapBufferEx is incorrect size");
struct IoctlUnmapBuffer {
s64_le offset{};
};
static_assert(sizeof(IoctlUnmapBuffer) == 8, "IoctlUnmapBuffer is incorrect size");
struct IoctlBindChannel {
s32_le fd{};
};
static_assert(sizeof(IoctlBindChannel) == 4, "IoctlBindChannel is incorrect size");
struct IoctlVaRegion {
u64_le offset{};
u32_le page_size{};
INSERT_PADDING_WORDS(1);
u64_le pages{};
};
static_assert(sizeof(IoctlVaRegion) == 24, "IoctlVaRegion is incorrect size");
struct IoctlGetVaRegions {
u64_le buf_addr{}; // (contained output user ptr on linux, ignored)
u32_le buf_size{}; // forced to 2*sizeof(struct va_region)
u32_le reserved{};
IoctlVaRegion small{};
IoctlVaRegion big{};
};
static_assert(sizeof(IoctlGetVaRegions) == 16 + sizeof(IoctlVaRegion) * 2,
"IoctlGetVaRegions is incorrect size");
s32 channel{};
u32 big_page_size{DEFAULT_BIG_PAGE_SIZE};
NvResult AllocAsEx(const std::vector<u8>& input, std::vector<u8>& output);
NvResult AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output);
NvResult Remap(const std::vector<u8>& input, std::vector<u8>& output);
NvResult MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output);
NvResult UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output);
NvResult FreeSpace(const std::vector<u8>& input, std::vector<u8>& output);
NvResult BindChannel(const std::vector<u8>& input, std::vector<u8>& output);
NvResult GetVARegions(const std::vector<u8>& input, std::vector<u8>& output);
NvResult GetVARegions(const std::vector<u8>& input, std::vector<u8>& output,
std::vector<u8>& inline_output);
std::optional<BufferMap> FindBufferMap(GPUVAddr gpu_addr) const;
void AddBufferMap(GPUVAddr gpu_addr, std::size_t size, VAddr cpu_addr, bool is_allocated);
std::optional<std::size_t> RemoveBufferMap(GPUVAddr gpu_addr);
Module& module;
NvCore::Container& container;
NvCore::NvMap& nvmap;
std::shared_ptr<Tegra::MemoryManager> gmmu;
// This is expected to be ordered, therefore we must use a map, not unordered_map
std::map<GPUVAddr, BufferMap> buffer_mappings;
};
} // namespace Service::Nvidia::Devices
|
